Abstract
Metagenomics-based studies have provided insight into many of the complex microbial communities responsible for maintaining life on this planet. Sequencing efforts often uncover novel genetic content; this is most evident for phage communities, in which upwards of 90% of all sequences exhibit no similarity to any sequence in current data repositories. For the small fraction that can be identified, the top BLAST hit is generally posited as being representative of a viral taxon present in the sample of origin. Homology-based classification, however, can be misleading as sequence repositories capture but a small fraction of phage diversity. Furthermore, lateral gene transfer is pervasive within phage communities. As such, the presence of a particular gene may not be indicative of the presence of a particular viral species. Rather, it is just that: an indication of the presence of a specific gene. To circumvent this limitation, we have developed a new method for the analysis of viral metagenomic datasets. BLAST hits are weighted, integrating the sequence identity and length of alignments as well as a taxonomic signal, such that each gene is evaluated with respect to its information content. Through this quantifiable metric, predictions of viral community structure can be made with confidence. As a proof-of-concept, the approach presented here was implemented and applied to seven freshwater viral metagenomes. While providing a robust method for evaluating viral metagenomic data, the tool is versatile and can easily be customized to investigations of any environment or biome.